Calibration method, device and equipment for vehicle-mounted laser radar

Abstract

This application provides a method, apparatus, and device for calibrating a vehicle-mounted laser radar. The method includes: performing a coarse calibration on the vehicle-mounted laser radar's raw point cloud; obtaining a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar based on the coarsely calibrated raw point cloud; obtaining roll and pitch parameters of the vehicle-mounted laser radar based on the ground point cloud; and obtaining yaw and altitude parameters of the vehicle-mounted laser radar based on the non-ground point cloud. Through the above method, this application can effectively improve the efficiency and accuracy of vehicle-mounted laser radar calibration.

Description

Technical Field

[0001] The present application relates to the field of sensor calibration technology, and in particular to a calibration method, device and equipment for a vehicle-mounted laser radar. Background Art

[0002] Automotive lidar is widely used in intelligent connected vehicles and self-driving cars. How to calibrate it quickly and accurately has always been a concern for researchers.

[0003] Accurate calibration of on-board LiDAR (LiDAR) is essential for autonomous vehicles to accurately perceive the external environment. The higher the calibration accuracy, the more accurately the point cloud data reflects the environment. Currently, LiDAR calibration is primarily achieved through methods such as recording a segment of data with the LiDAR and then performing offline parameter calculations using a point cloud registration algorithm, or by parking the vehicle in a specific position and then calibrating the LiDAR using calibration objects. These methods are cumbersome and complex, requiring specialized personnel to perform, resulting in low accuracy and efficiency in LiDAR external parameter calibration, making them difficult to meet the calibration requirements of mass-produced autonomous vehicles. Summary of the Invention

[0004] In view of the above problems, the present application provides a calibration method, device and equipment for a vehicle-mounted laser radar to solve the problems of low efficiency and accuracy of current vehicle-mounted laser radar calibration methods.

[0005] In order to achieve the above objectives, this application provides the following technical solutions:

[0006] According to a first aspect of the present application, a calibration method for a vehicle-mounted laser radar is provided, comprising:

[0007] Roughly calibrate the original point cloud of the vehicle-mounted lidar;

[0008] Acquire the ground point cloud and non-ground point cloud corresponding to the vehicle-mounted laser radar based on the original point cloud after rough calibration;

[0009] Acquire a roll angle parameter and a pitch angle parameter of the vehicle-mounted laser radar based on the ground point cloud;

[0010] The yaw angle parameter and the altitude parameter of the vehicle-mounted laser radar are obtained based on the non-ground point cloud.

[0011] In one embodiment, obtaining the ground point cloud and non-ground point cloud corresponding to the vehicle-mounted laser radar based on the raw point cloud after rough calibration includes:

[0012] The raw point cloud after rough calibration is subjected to a first processing to obtain a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar; wherein the first processing includes downsampling processing and ground removal processing.

[0013] In one embodiment, the first processing of the coarsely calibrated original point cloud includes:

[0014] Downsample the original point cloud after coarse calibration;

[0015] Rasterize the downsampled point cloud and calculate the maximum height difference between all points in each grid and the point cloud within the preset range;

[0016] If the maximum height difference is greater than a preset threshold, all points in the grid are recorded as non-ground point clouds;

[0017] If the maximum height difference is not greater than a preset threshold, all points in the grid are recorded as ground point clouds.

[0018] In one embodiment, the acquiring of the roll angle parameter and the pitch angle parameter of the vehicle-mounted laser radar based on the ground point cloud includes:

[0019] Performing plane fitting on the ground point cloud and obtaining a normal vector of the ground point cloud after plane fitting;

[0020] The roll angle parameter and pitch angle parameter of the vehicle-mounted laser radar are obtained based on the normal vector.

[0021] In one embodiment, performing plane fitting on the ground point cloud and obtaining a normal vector of the ground point cloud after plane fitting includes:

[0022] Setting a plurality of first regions of interest in the ground point cloud;

[0023] randomly selecting point clouds from the plurality of first regions of interest, respectively, and estimating a reference equation for plane fitting based on the point clouds;

[0024] Obtaining the number of point clouds within the range of the benchmark equation;

[0025] If the maximum number of iterations is not reached, returning to the step of randomly selecting point clouds from the plurality of first regions of interest;

[0026] If the maximum number of iterations is reached, the benchmark equation with the largest number of point clouds within the benchmark equation range is selected from the benchmark equations obtained in each iteration, and the normal vector of the ground point cloud is obtained based on the benchmark equation with the largest number of point clouds.

[0027] In one embodiment, the acquiring of the yaw angle parameter and the altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud includes:

[0028] Rotating the non-ground point cloud based on a first extrinsic parameter, where the first extrinsic parameter includes an initial yaw angle parameter, a roll angle parameter, and a pitch angle parameter for coarsely calibrating the original point cloud;

[0029] Segmenting the rotated non-ground point cloud based on a plurality of second regions of interest to obtain a vehicle point cloud;

[0030] Converting the vehicle point cloud into an image, and performing a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation;

[0031] Obtaining a yaw angle parameter of the laser radar based on a vehicle profile and a corresponding orientation;

[0032] The average height of the second area of interest point cloud is calculated based on the second external parameter to obtain the height parameter of the vehicle-mounted laser radar, where the second external parameter includes the roll angle parameter, the pitch angle parameter and the yaw angle parameter.

[0033] In one embodiment, converting the vehicle point cloud into an image and performing a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation includes:

[0034] Projecting the vehicle point cloud onto the XY plane of the laser radar coordinate system to generate a bird's-eye view image;

[0035] Performing convex hull calculation on the two-dimensional point cloud in the bird's-eye view image based on a convex hull calculation function to obtain polygons corresponding to multiple corner points of the vehicle;

[0036] Calculate the length of each side of the polygon and the height of its corresponding bounding box respectively;

[0037] A bounding box with the smallest area is selected from the bounding boxes corresponding to each edge as the outline of the vehicle, and a longer direction of the length of the edge and the height corresponding to the outline is selected as the vehicle orientation.

[0038] In one embodiment, before coarsely calibrating the original point cloud of the vehicle-mounted laser radar, the method further includes:

[0039] Select a horizontal road that meets the preset conditions so that the vehicle-mounted lidar can collect the vehicle's original point cloud;

[0040] Obtain the original point cloud collected by the vehicle-mounted laser radar.

[0041] According to a second aspect of the present application, a calibration device for a vehicle-mounted laser radar is provided, comprising:

[0042] A coarse calibration module, configured to perform coarse calibration on the original point cloud of the vehicle-mounted lidar;

[0043] An acquisition module configured to acquire a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar based on the original point cloud after rough calibration;

[0044] a first fine calibration module, configured to obtain a roll angle parameter and a pitch angle parameter of the vehicle-mounted laser radar based on the ground point cloud;

[0045] The second fine calibration module obtains the yaw angle parameter and altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud.

[0046] According to a third aspect of the present application, there is provided an electronic device, comprising: a memory and a processor;

[0047] The memory stores computer-executable instructions;

[0048] The processor executes the computer-executable instructions stored in the memory, so that the electronic device executes the calibration method of the vehicle-mounted laser radar.

[0049] It can be understood that the calibration method, device and equipment of the vehicle-mounted laser radar provided in this application perform rough calibration on the external parameters of the laser radar, downsample the original point cloud that has undergone rough calibration, segment the ground, extract the road point cloud and the vehicle's own point cloud, fit the horizontal road surface and vehicle orientation, so as to calculate the external parameters of the vehicle-mounted laser radar. The calibration process is simple and efficient, and the operation flow is simple. It can automatically and efficiently calibrate the laser radar, effectively improving the efficiency and accuracy of the vehicle-mounted laser radar calibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0051] Figure 1 A schematic diagram of a possible scenario provided in an embodiment of the present application;

[0052] Figure 2 A schematic diagram of a flow chart of a method for calibrating a vehicle-mounted laser radar provided in an embodiment of the present application;

[0053] Figure 3 This is a schematic diagram of the installation position of the vehicle-mounted laser radar in the embodiment of the present application;

[0054] Figure 4 A schematic diagram of a region of interest of a fitted plane in an embodiment of the present application;

[0055] Figure 5This is a schematic diagram of fitting the vehicle body contour and orientation in an embodiment of the present application;

[0056] Figure 6 A schematic diagram of the structure of a calibration device for a vehicle-mounted laser radar provided in an embodiment of the present application;

[0057] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0058] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0059] Before introducing the embodiments of the present application, the technical background of the embodiments of the present application is first explained: In the related art, the laser radar module in the autonomous driving vehicle perception system is composed of one or more vehicle-mounted laser radars to perceive the environment around the vehicle body. However, in order to fuse the data between multiple laser radars or between the laser radar and other sensors, it is necessary to uniformly convert the data of the laser radar and other sensors into the same coordinate system, generally into the vehicle body coordinate system. At this time, it is necessary to calibrate the rotation and translation relationship between the coordinate system of each laser radar and the vehicle body coordinate system, which can be represented by the rotation matrix R and the translation vector t. However, the current laser radar calibration steps are cumbersome and complicated, and require professional personnel to operate. Not only has there been no standardized and automated calibration method, but the accuracy and efficiency of manual calibration are low, and it cannot meet the laser radar calibration work requirements of mass-produced autonomous driving vehicles.

[0060] In response to the above technical problems, the embodiments of the present application provide a calibration method, device and equipment for a vehicle-mounted laser radar. By coarsely calibrating the external parameters of the laser radar, downsampling the original point cloud that has undergone coarse calibration, segmenting the ground, extracting the road point cloud and the vehicle's own point cloud, fitting the ground point cloud and the vehicle orientation, the external parameters of the vehicle-mounted laser radar are calculated. The calibration process is simple and efficient, with a simple operation flow. It can automatically and efficiently calibrate the laser radar, effectively improving the efficiency and accuracy of the vehicle-mounted laser radar calibration.

[0061] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described in more detail below in conjunction with the drawings in the embodiments of the present application. In the drawings, the same or similar reference numerals throughout represent the same or similar parts or parts with the same or similar functions. The described embodiments are part of the embodiments of the present application, not all of the embodiments. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain the present application, and should not be understood as limitations on the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0062] Figure 1 A possible scenario diagram provided for an embodiment of the present application is as follows: Figure 1 As shown, the system includes a server 110 and a vehicle-mounted laser radar 120, which are interconnected via a wired or wireless network. In some embodiments, the vehicle-mounted laser radar 120 is used to provide the server 110 with the raw point cloud data it has collected, and the server 110 is used to perform extrinsic parameter calibration on the raw point cloud data provided by the vehicle-mounted laser radar 120. Optionally, there can be one or more vehicle-mounted laser radars 120, and during the extrinsic parameter calibration process, the server 110 can undertake the main computing work or all computing work.

[0063] Among them, server 110 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0064] The vehicle-mounted LiDAR 120, also known as a vehicle-mounted 3D laser scanner, can be used to collect point cloud data. It is a mobile 3D laser scanning system and one of the most effective tools for urban modeling. In order to fuse data between multiple LiDARs or between LiDARs and other sensors, the data from LiDARs and other sensors needs to be uniformly converted to the same coordinate system, generally to the vehicle coordinate system. At this time, the rotation and translation relationship between the coordinate system of each LiDAR and the vehicle coordinate system needs to be calibrated.

[0065] Optionally, the number of the above-mentioned servers 110 or vehicle-mounted laser radars 120 can be more or less, and this embodiment of the application does not limit this. In some embodiments, the above-mentioned servers 110 and vehicle-mounted laser radars 120 can be installed in the same smart vehicle.

[0066] The above is a brief description of the scenario diagram of this application. Figure 1 Taking the server 120 in the example as an example, the calibration method of the vehicle-mounted laser radar provided in the embodiment of the present application is described in detail.

[0067] Please refer to Figure 2 , Figure 2 For the embodiment of the present application, according to the first aspect of the present application, a flow chart of a calibration method of a vehicle-mounted laser radar is provided, and the method includes steps S201-S204.

[0068] Step S201: perform coarse calibration on the original point cloud of the vehicle-mounted laser radar.

[0069] In one possible implementation, the rough calibration method can be to use the on-board laser radar installed on the left front of the vehicle to collect a point cloud cluster of a vehicle traveling at a constant speed. The on-board laser radar in this example takes the Velodyne 32-line laser radar as an example, with a vertical scanning angle of 30 degrees and a horizontal scanning angle of 360 degrees, and a rotation speed of 10HZ. The specific installation method is that the X-axis is 45 degrees to the left of the vehicle's driving direction, the Y-axis is 135 degrees to the left of the vehicle's driving direction, the Z-axis is perpendicular to the horizontal plane where the vehicle is located, and the coordinate origin is the intersection of the center of the vehicle's rear axle and the ground plane. Among them, the installation diagram of the on-board laser radar is as follows Figure 3 As shown, the left part is a front view and the right part is a top view.

[0070] The original point cloud cluster collected by the vehicle-mounted lidar is a point set containing three coordinates: X, Y, and Z (vehicle-mounted lidar coordinate system). The collected point cloud is roughly estimated by manual measurement of the lidar's attitude angle relative to the road surface, including heading angle, pitch angle, roll angle, etc. These estimated parameters are used as the initial parameters for calibration. With reference to this initial parameter, the original point cloud can be projected into the real coordinate system (vehicle body coordinate system).

[0071] Furthermore, in order to improve the accuracy of the vehicle-mounted LiDAR calibration, the accuracy of the original point cloud is taken into consideration when collecting the original point cloud. Specifically, before the rough calibration of the original point cloud of the vehicle-mounted LiDAR, the following steps are also included:

[0072] A horizontal road that meets preset conditions is selected so that the vehicle-mounted laser radar can collect the original point cloud of the vehicle; and the original point cloud collected by the vehicle-mounted laser radar is obtained.

[0073] For example, a horizontal road with a length of 30 meters and a width of more than 10 meters can be selected, and there are no other obstacles on the site. Then, a laser radar installed in a fixed position on the vehicle is used to collect a section of original point cloud.

[0074] It should be noted that those skilled in the art can adaptively set the preset conditions in combination with actual applications. For example, a horizontal road without obstacle interference can be selected, that is, there are no other obstacles on the horizontal road except the vehicle, and the on-board lidar can detect the side profile of the vehicle to ensure that the on-board lidar can collect the original point cloud of the vehicle, and try to eliminate the interference factors of obstacles in the original point cloud of the vehicle collected by the on-board lidar to improve the calibration accuracy of the lidar.

[0075] Step S202: Acquire the ground point cloud and non-ground point cloud corresponding to the vehicle-mounted laser radar based on the original point cloud after rough calibration.

[0076] In this embodiment, the original point cloud after coarse calibration is used to perform downsampling and ground removal operations in sequence to obtain the ground point cloud and non-ground point cloud corresponding to the vehicle-mounted laser radar, and then the corresponding laser radar external parameters are obtained based on the ground point cloud and non-ground point cloud respectively.

[0077] In one embodiment, the specific process of obtaining the ground point cloud and the non-ground point cloud is as follows:

[0078] The raw point cloud after rough calibration is subjected to a first processing to obtain a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar; wherein the first processing includes downsampling processing and ground removal processing.

[0079] The original point cloud is downsampled to remove noise point clouds and improve computing speed, while the ground removal process can distinguish ground point clouds from non-ground point clouds. In a further embodiment, the raw point cloud after rough calibration is first processed, including the following steps:

[0080] Downsample the original point cloud after coarse calibration;

[0081] Rasterize the downsampled point cloud and calculate the maximum height difference between all points in each grid and the point cloud within the preset range;

[0082] If the maximum height difference is greater than a preset threshold, all points in the grid are recorded as non-ground point clouds;

[0083] If the maximum height difference is not greater than a preset threshold, all points in the grid are recorded as ground point clouds.

[0084] In this embodiment, by rasterizing the downsampled point cloud, calculating the height difference of the points in the rasterization, and then dividing the point cloud in the grid into ground point cloud or non-ground point cloud, the acquisition efficiency of ground point cloud and non-ground point cloud can be effectively improved.

[0085] Step S203: Obtain roll angle parameters and pitch angle parameters of the vehicle-mounted laser radar based on the ground point cloud.

[0086] Specifically, plane fitting is performed on the ground point cloud to obtain the roll angle parameters and pitch parameters of the vehicle-mounted lidar.

[0087] In one embodiment, step S203 specifically includes the following steps: performing plane fitting on the ground point cloud and obtaining the normal vector of the ground point cloud after plane fitting; and obtaining the roll angle parameters and pitch angle parameters of the vehicle-mounted laser radar based on the normal vector.

[0088] Specifically, the ground point cloud can be fitted with a plane and the reference equation of the plane-fitted ground point cloud can be calculated. The direction is obtained based on the reference equation, and the angle between the normal vector and the standard normal vector is the lidar roll angle and pitch angle. It is understood that the standard vector can be the normal vector of the real ground plane.

[0089] It can be understood that plane fitting is a relatively mature existing technology, and this embodiment will not elaborate on it in detail.

[0090] In a further embodiment, performing plane fitting on the ground point cloud and obtaining a normal vector of the ground point cloud after plane fitting includes:

[0091] Setting a plurality of first regions of interest in the ground point cloud;

[0092] randomly selecting point clouds from the plurality of first regions of interest, respectively, and estimating a reference equation for plane fitting based on the point clouds;

[0093] Obtaining the number of point clouds within the range of the benchmark equation;

[0094] If the maximum number of iterations is not reached, returning to the step of randomly selecting point clouds from the plurality of first regions of interest;

[0095] If the maximum number of iterations is reached, the benchmark equation with the largest number of point clouds within the benchmark equation range is selected from the benchmark equations obtained in each iteration, and the normal vector of the ground point cloud is obtained based on the benchmark equation with the largest number of point clouds.

[0096] In one example, a) four ROI regions of interest (ROIs) are set in the ground point cloud, and all ground point clouds within the above regions of interest are extracted to fit the ground point cloud plane equation. The four regions of interest in the ground fitting are shown in the figure below. Figure 4As shown; b) randomly select three point clouds in the area of interest and estimate the reference equation of the road surface (plane fitting) ax+by+cz+d=0; c) analyze the ground point cloud within the above reference equation, record the equation and the number of point clouds within it; repeat steps b) and c) until the number of iterations is exceeded. During this period, the plane equation with the largest number of point clouds within the reference equation is the reference equation of the road surface, and the normal vector of the ground point cloud can be obtained. Among them, the fitted plane equation can obtain the ground normal vector VA=(a,b,c).

[0097] Step S204: Acquire the yaw angle parameter and altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud.

[0098] In this embodiment, the vehicle body orientation is fitted to the non-ground point cloud, that is, the outline and orientation of the vehicle are fitted using the non-ground point cloud. Then, the yaw angle can be calculated by calculating the angle between the vehicle orientation and the X-axis of the lidar installation position, and the height parameter of the lidar can be obtained by calculating the average height of the point cloud.

[0099] In one embodiment, step S204 specifically includes the following steps:

[0100] Rotating the non-ground point cloud based on a first extrinsic parameter, where the first extrinsic parameter includes an initial yaw angle parameter, a roll angle parameter, and a pitch angle parameter for coarsely calibrating the original point cloud;

[0101] Segmenting the rotated non-ground point cloud based on a plurality of second regions of interest to obtain a vehicle point cloud;

[0102] Converting the vehicle point cloud into an image, and performing a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation;

[0103] Obtaining a yaw angle parameter of the laser radar based on a vehicle profile and a corresponding orientation;

[0104] The average height of the second area of interest point cloud is calculated based on the second external parameter to obtain the height parameter of the vehicle-mounted laser radar, where the second external parameter includes the roll angle parameter, the pitch angle parameter and the yaw angle parameter.

[0105] Specifically, the first external parameter is the current external parameter, which includes the initial external parameter and the calibrated external parameter. The calibrated external parameter includes the roll angle parameter and the pitch angle parameter. The calibrated external parameter is used to replace the initial roll angle parameter and the pitch angle parameter in the initial external parameter, and the initial yaw angle parameter in the coarse calibration that has not yet been finely calibrated is retained to rotate the non-ground point cloud. Specifically, the first external parameter is used to calculate the rotation matrix, and then the non-ground point cloud is rotated according to the rotation matrix.

[0106] Specifically, four regions of interest (ROIs) can also be set to segment the non-ground point cloud to obtain the vehicle's own point cloud. It should be noted that the first and second ROIs in this embodiment are only used to distinguish similar objects and have no other meaning. In some embodiments, ROIs can be set first, and then the ground point cloud and non-ground point cloud can be segmented simultaneously based on the ROIs.

[0107] Specifically, the vehicle orientation obtained by fitting is VD = (x, y, 0), the vehicle orientation in the vehicle body coordinate system is the X-axis, and the angle between the X-axis of the lidar and the X-axis of the vehicle can be obtained, which is the heading angle.

[0108] Specifically, the first extrinsic parameter and the second extrinsic parameter are the latest calibration extrinsic parameters. The second extrinsic parameter is further updated (yaw angle parameter) compared to the first extrinsic parameter. The original point cloud is calibrated with the obtained new extrinsic parameter, and the average height of the point cloud in the area of interest is calculated to obtain the lidar height parameter.

[0109] In one embodiment, converting the vehicle point cloud into an image and performing a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation includes the following steps:

[0110] Projecting the vehicle point cloud onto the XY plane of the laser radar coordinate system to generate a bird's-eye view image;

[0111] Performing convex hull calculation on the two-dimensional point cloud in the bird's-eye view image based on a convex hull calculation function to obtain polygons corresponding to multiple corner points of the vehicle;

[0112] Calculate the length of each side of the polygon and the height of its corresponding bounding box respectively;

[0113] A bounding box with the smallest area is selected from the bounding boxes corresponding to each edge as the outline of the vehicle, and a longer direction of the length of the edge and the height corresponding to the outline is selected as the vehicle orientation.

[0114] Specifically, the vehicle's own point cloud is projected onto the XY plane to generate a bird's-eye view image, and the image is processed using image morphology methods to obtain clear edges. The convex hull calculation function in the computer vision library is then called to perform convex hull calculation on the two-dimensional point cloud corresponding to the bird's-eye view image to obtain the corner points of the vehicle polygon. One edge of the polygon is selected, and all the remaining points are projected onto this edge. The distance between the two farthest projection points is calculated as the length of this edge, and the point farthest from the edge is selected to calculate the height of the Box. Finally, the bounding box with the smallest area is selected as the bounding box of the vehicle. The lengths of the edge and height are compared, and the longer one is the orientation of the vehicle. The fitted vehicle outline and the corresponding orientation are as follows: Figure 5 shown.

[0115] In summary, the laser radar calibration method of this embodiment mainly includes the following steps: the vehicle-mounted laser radar collects the original point cloud; performs coarse calibration on the original point cloud; downsamples the original point cloud; performs ground segmentation on the original point cloud; performs ground fitting on the ground point cloud; performs vehicle body orientation fitting on the ground point cloud; and finally calculates the laser radar extrinsic parameters based on the ground fitting results and the vehicle body orientation fitting results.

[0116] Compared with the related art, the calibration steps of vehicle-mounted LiDAR are cumbersome and complicated, and it is impossible to standardize and automate the efficient and accurate calibration of LiDAR. In this embodiment, the LiDAR extrinsic parameters are roughly calibrated. By downsampling the coarsely calibrated original point cloud and segmenting the ground, the road point cloud and the vehicle point cloud are extracted, the ground point cloud and the vehicle's orientation are fitted, and the vehicle-mounted LiDAR extrinsic parameters are calculated. The calibration process is simple, efficient, and has a simple operation flow. It can automatically and efficiently calibrate the LiDAR, effectively improving the efficiency and accuracy of vehicle-mounted LiDAR calibration.

[0117] In addition, by selecting a horizontal road without obstacle interference, the original point cloud of the vehicle-mounted lidar is obtained to reduce the influence of obstacles; and the lidar roll angle and pitch angle are calculated using the fitting equation of the horizontal road surface. At the same time, the imaging method is used to fit the vehicle's own orientation to calculate the lidar yaw angle to obtain the vehicle-mounted lidar external parameters, which can further improve the calibration efficiency and accuracy of the lidar external parameters.

[0118] According to the second aspect of the embodiment of the present application, a calibration device for a vehicle-mounted laser radar is also provided. Figure 6 Shown, including:

[0119] A coarse calibration module 61 is configured to perform coarse calibration on the original point cloud of the vehicle-mounted laser radar;

[0120] An acquisition module 62 is configured to acquire a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar based on the original point cloud after rough calibration;

[0121] A first fine calibration module 63 is configured to obtain roll angle parameters and pitch angle parameters of the vehicle-mounted laser radar based on the ground point cloud;

[0122] The second fine calibration module 64 obtains the yaw angle parameter and the altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud.

[0123] In one embodiment, the acquisition module 62 includes:

[0124] The first processing unit is configured to perform a first processing on the original point cloud after rough calibration to obtain a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar; wherein the first processing includes downsampling processing and ground removal processing.

[0125] In one embodiment, the first processing unit is specifically configured to downsample the original point cloud after coarse calibration; rasterize the downsampled point cloud, and calculate the maximum height difference between all points in each grid and the point cloud within a preset range; if the maximum height difference is greater than a preset threshold, all points in the grid are recorded as non-ground point clouds; if the maximum height difference is not greater than the preset threshold, all points in the grid are recorded as ground point clouds.

[0126] In one embodiment, the first fine calibration module includes:

[0127] a normal vector acquisition unit, configured to perform plane fitting on the ground point cloud and acquire a normal vector of the ground point cloud after plane fitting;

[0128] The first parameter acquisition unit is configured to acquire the roll angle parameter and the pitch angle parameter of the vehicle-mounted laser radar based on the normal vector.

[0129] In one embodiment, the normal vector acquisition unit is specifically configured to set several first regions of interest in the ground point cloud that has undergone plane fitting; randomly select point clouds from the several first regions of interest respectively, and estimate the reference equation of the road surface based on the point clouds; obtain the number of point clouds within the range of the reference equation; if the maximum number of iterations is not reached, return to the step of randomly selecting point clouds from the several first regions of interest respectively; if the maximum number of iterations is reached, select the reference equation with the largest number of point clouds within the range of the reference equation from the reference equations obtained in each iteration, and obtain the normal vector of the ground point cloud based on the reference equation with the largest number of point clouds.

[0130] In one embodiment, the second fine calibration module includes:

[0131] a rotation unit configured to rotate the non-ground point cloud based on a first extrinsic parameter, wherein the first extrinsic parameter includes an initial yaw angle parameter for coarsely calibrating the original point cloud, the roll angle parameter, and the pitch angle parameter;

[0132] a segmentation unit configured to segment the rotated non-ground point cloud based on a plurality of second regions of interest to obtain a vehicle point cloud;

[0133] a second processing unit configured to convert the vehicle point cloud into an image, and perform a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation;

[0134] a second parameter acquisition unit configured to acquire a yaw angle parameter of the laser radar based on a vehicle profile and a corresponding orientation;

[0135] A third parameter acquisition unit is configured to calculate the average height of the second area of interest point cloud based on a second external parameter to obtain a height parameter of the vehicle-mounted laser radar, where the second external parameter includes the roll angle parameter, the pitch angle parameter, and the yaw angle parameter.

[0136] In one embodiment, the second processing unit is specifically configured to project the vehicle point cloud onto the XY plane of the coordinate system of the laser radar to generate a bird's-eye view image; perform convex hull calculation on the two-dimensional point cloud in the bird's-eye view image based on a convex hull calculation function to obtain polygons corresponding to multiple corner points of the vehicle; calculate the length of each side of the polygon and the height of its corresponding bounding box respectively; select the bounding box with the smallest area from the bounding boxes corresponding to each side as the outline of the vehicle, and select the longer direction of the length of the side and the height corresponding to the outline as the vehicle orientation.

[0137] In one embodiment, the apparatus further comprises:

[0138] A selection module is configured to select a horizontal road that meets preset conditions so that the vehicle-mounted laser radar can collect an original point cloud of the vehicle;

[0139] The point cloud acquisition module is configured to acquire the original point cloud collected by the vehicle-mounted laser radar.

[0140] It should be noted here that the calibration device of the above-mentioned vehicle-mounted laser radar provided in this application can implement all the method steps implemented in the above-mentioned method embodiment and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0141] According to a third aspect of the embodiments of the present application, an electronic device is also provided, such as Figure 7 As shown, it includes: a memory 71 and a processor 72;

[0142] The memory 71 stores computer-executable instructions;

[0143] The processor 72 executes the computer-executable instructions stored in the memory 71, so that the electronic device executes the calibration method of the vehicle-mounted laser radar.

[0144] It should be noted here that the above-mentioned electronic device provided in this application can implement all the method steps implemented in the above-mentioned method embodiment and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0145] According to the fourth aspect of the embodiment of the present application, a processor-readable storage medium is also provided, wherein the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the calibration method of the vehicle-mounted laser radar.

[0146] It should be noted here that when the processor executes the computer program, it can implement all the method steps implemented by the network device in the above method embodiment and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0147] It will be understood by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium).

[0148] As is well known to those skilled in the art, the term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer.

[0149] Furthermore, as is well known to those skilled in the art, communication media typically embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0150] In the description of the embodiments of the present application, the term "and / or" merely represents an association relationship that describes associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" represents any combination of at least two of any one or more of a plurality of items. For example, at least one of A, B, and C can represent any one or more elements selected from a set that includes A, B, and C. In addition, the term "plurality" means two or more, unless otherwise specified.

[0151] In the description of the embodiments of the present application, the terms "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A calibration method for a vehicle-mounted laser radar, characterized in that: include: Roughly calibrate the original point cloud of the vehicle-mounted lidar; Acquire the ground point cloud and non-ground point cloud corresponding to the vehicle-mounted laser radar based on the original point cloud after rough calibration; Acquire a roll angle parameter and a pitch angle parameter of the vehicle-mounted laser radar based on the ground point cloud; Acquire a yaw angle parameter and an altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud; The obtaining of the yaw angle parameter and the altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud includes: Rotating the non-ground point cloud based on a first extrinsic parameter, where the first extrinsic parameter includes an initial yaw angle parameter, a roll angle parameter, and a pitch angle parameter for coarsely calibrating the original point cloud; Segmenting the rotated non-ground point cloud based on a plurality of second regions of interest to obtain a vehicle point cloud; Converting the vehicle point cloud into an image, and performing a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation; Obtaining a yaw angle parameter of the laser radar based on a vehicle profile and a corresponding orientation; The average height of the second area of interest point cloud is calculated based on the second external parameter to obtain the height parameter of the vehicle-mounted laser radar, where the second external parameter includes the roll angle parameter, the pitch angle parameter and the yaw angle parameter.

2. The method according to claim 1, characterized in that The obtaining of the ground point cloud and the non-ground point cloud corresponding to the vehicle-mounted laser radar based on the rough-calibrated original point cloud includes: The raw point cloud after rough calibration is subjected to a first processing to obtain a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar; wherein the first processing includes downsampling processing and ground removal processing.

3. The method according to claim 2, characterized in that The first processing of the rough-calibrated original point cloud includes: Downsample the original point cloud after coarse calibration; Rasterize the downsampled point cloud and calculate the maximum height difference between all points in each grid and the point cloud within the preset range; If the maximum height difference is greater than a preset threshold, all points in the grid are recorded as non-ground point clouds; If the maximum height difference is not greater than a preset threshold, all points in the grid are recorded as ground point clouds.

4. The method according to any one of claims 1 to 3, characterized in that The acquiring of the roll angle parameter and the pitch angle parameter of the vehicle-mounted laser radar based on the ground point cloud includes: Performing plane fitting on the ground point cloud and obtaining a normal vector of the ground point cloud after plane fitting; The roll angle parameter and pitch angle parameter of the vehicle-mounted laser radar are obtained based on the normal vector.

5. The method according to claim 4, characterized in that The performing plane fitting on the ground point cloud and obtaining a normal vector of the ground point cloud after plane fitting includes: Setting a plurality of first regions of interest in the ground point cloud; randomly selecting point clouds from the plurality of first regions of interest, respectively, and estimating a reference equation for plane fitting based on the point clouds; Obtaining the number of point clouds within the range of the benchmark equation; If the maximum number of iterations is not reached, returning to the step of randomly selecting point clouds from the plurality of first regions of interest; If the maximum number of iterations is reached, the benchmark equation with the largest number of point clouds within the benchmark equation range is selected from the benchmark equations obtained in each iteration, and the normal vector of the ground point cloud is obtained based on the benchmark equation with the largest number of point clouds.

6. The method according to claim 1, characterized in that The converting the vehicle point cloud into an image and performing a second processing on the image based on an image processing algorithm to obtain a vehicle outline and a corresponding orientation includes: Projecting the vehicle point cloud onto the XY plane of the laser radar coordinate system to generate a bird's-eye view image; Performing convex hull calculation on the two-dimensional point cloud in the bird's-eye view image based on a convex hull calculation function to obtain polygons corresponding to multiple corner points of the vehicle; Calculate the length of each side of the polygon and the height of its corresponding bounding box respectively; A bounding box with the smallest area is selected from the bounding boxes corresponding to each edge as the outline of the vehicle, and a longer direction of the length of the edge and the height corresponding to the outline is selected as the vehicle orientation.

7. The method according to claim 1, characterized in that Before coarse calibration of the original point cloud of the vehicle-mounted lidar, it also includes: Select a horizontal road that meets the preset conditions so that the vehicle-mounted lidar can collect the vehicle's original point cloud; Obtain the original point cloud collected by the vehicle-mounted laser radar.

8. A calibration device for a vehicle-mounted laser radar, characterized in that: include: A coarse calibration module, configured to perform coarse calibration on the original point cloud of the vehicle-mounted lidar; An acquisition module, configured to acquire a ground point cloud and a non-ground point cloud corresponding to the vehicle-mounted laser radar based on the original point cloud after rough calibration; a first fine calibration module, configured to obtain a roll angle parameter and a pitch angle parameter of the vehicle-mounted laser radar based on the ground point cloud; A second fine calibration module obtains a yaw angle parameter and an altitude parameter of the vehicle-mounted laser radar based on the non-ground point cloud; The second fine calibration module is specifically used to rotate the non-ground point cloud based on a first external parameter, wherein the first external parameter includes an initial yaw angle parameter for coarsely calibrating the original point cloud, as well as the roll angle parameter and the pitch angle parameter; segment the rotated non-ground point cloud based on a number of second regions of interest to obtain a vehicle point cloud; convert the vehicle point cloud into an image, and perform a second processing on the image based on an image processing algorithm to obtain a vehicle contour and a corresponding orientation; obtain the yaw angle parameter of the laser radar based on the vehicle contour and the corresponding orientation; calculate the average height of the point cloud of the second region of interest based on a second external parameter to obtain a height parameter of the vehicle-mounted laser radar, wherein the second external parameter includes the roll angle parameter, the pitch angle parameter and the yaw angle parameter.

9. An electronic device, characterized in that: include: memory and processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the electronic device executes the calibration method of the vehicle-mounted laser radar according to any one of claims 1 to 7.

Images